Arnold Kling  

Daniel Kahneman's Thinking

PRINT
Internships vs. Apprenticeship... Abba Lerner on Consumer Sovere...

The book is called Thinking, Fast and Slow and for me it is one of the best five books of the year. No, he is not attempting to break new ground. It is more like a summing up of his career and of the topic of behavioral economics. But even if it is mostly familiar to you, I still recommend it. If it is less familiar to you, I recommend it even more.

Apropos my discussion of Super-achievers, I liked this paragraph on page 256:


Optimistic individuals play a disproportionate role in shaping our lives...they are the inventors, the entrepreneurs, the political and military leaders--not average people. They got to where they are by seeking challenges and taking risks. They are talented and they have been lucky, almost certainly luckier than they acknowledge. They are probably optimistic by temperament; a survey of founders of small businesses concluded that entrepreneurs are more sanguine than midlevel managers about life in general. Their experiences of success have confirmed their faith in their judgment and in their ability to control events. Their self-confidence is reinforced by the admiration of others. This reasoning leads to a hypothesis: the people who have the greatest influence on the lives of others are likely to be optimistic and overconfident, and to take more risks than they realize.

What bothers me about behavioral economists is what I might term "one-trial bias." Most of the results (not all) are from situations in which individuals are confronted with a problem that is novel for them. Often, the psychologist-experimenter engages in deception. The mistakes that individuals make may or may not be replicated in repeated situations, in circumstances where the individual is able to learn from experience, or in situations where an organizational or institutional mechanism (such as the market) may produce results that are superior to the judgment of any single individual.

Kahneman's insistence that success involves a great deal of luck (he makes this point several times in the book) may also be a case of "one-trial bias." In this essay, I argue that it is important to remember that a game lasts more than one move. Two chess players, for example, may appear to be so close in terms of ability that it would seem that the outcome should be determined by luck. If you look at any one move, one player may have a 70 percent chance of making the optimal move and the other player may have a 65 percent chance. If that one move were the whole game, then the outcome would vary considerably, depending on luck. However, in a sequence of many moves, the better player's chance of coming out ahead gets to be very high. Business is like a multi-move game, in which any one decision by an inferior player can turn out right, but it is unlikely that an inferior player will make a sequence of decisions that turns out to be better than those of a superior player.

I certainly believe that the phenomenon of overconfidence is important. I believe that hindsight bias and the fundamental attribution error are at work in leading people to over-estimate the skill of CEO's. However, in some places in the book, Kahneman assigns numerical values to the percentage of success that is due to luck that are purely subjective. That is, they are not based on any study, but instead represent his gut feel. I worry that his gut feel may fail to take into account the context of multiple decisions, learning, and the ability of the market to sort for skill.

On p. 412, he writes,


For behavioral economists, however, freedom has a cost, which is borne by individuals who make bad choices, and by a society that feels obligated to help them. The decision of whether to protect individuals against their mistakes therefore presents a dilemma for behavioral economists.

This poses a choice between freedom, with mistakes, and state power. I think of this as a false choice. As I wrote in The Era of Expert Failure, the choice we face is not between following government experts or making our own mistakes. Instead, it is a choice between following the experts who emerge in a competitive market or obeying experts who grasp the reins of power in government. Government is not the only source, or even the best source, of restraints on our propensity to make mistakes. The market provides many such mechanisms. I would rather see behavioral economists attempting to be market entrepreneurs than see them acting as policy entrepreneurs.

Do not let my critical comments turn you away from the book. It is rich with insight. One example is the way that statistical prediction often beats clinical prediction. But there are many other examples.



COMMENTS (10 to date)
Blake writes:

If the problem isn't really with experts per se then why the focus on expert failure?
If your argument boils back down to a problem with experts when linked with power and you're really just making a pretty standard 'we should disperse power' argument, then what is there to say about experts? Experts, like everyone, are better when they have proper market incentives. The end.

Is it because we sometimes think experts don't need proper incentives or that it is perhaps in the nature of expertise to already have the correct incentives for truth. After all, unpaid doctors still do their best don't they?

Adam writes:

Blake: it's not just about the incentives you give experts. Markets actually provide a process for filtering out who is an expert in the first place. The feedback mechanisms for who gets pronounced an expert in the market vs. who gets pronounced an expert in government are very different, with the latter being political rather than merit-based.

Becky Hargrove writes:

I would enjoy the book, but not ascribe as much importance to overconfidence and luck in success. Sometimes success is subjective in simply knowing one has finally taken the right path.

However, I think that behavioral economists have not quite come to the calling that could matter the most, which is to reengage people in the importance of economic life. The focus on policy effects is akin to the focus on the supposedly passive interests of the consumer which is utter nonsense. Our identities are completely caught up in the degree of economic access we have in our surroundings and I would love to see more behavioral studies in that direction.

GU writes:
"Business is like a multi-move game, in which any one decision by an inferior player can turn out right, but it is unlikely that an inferior player will make a sequence of decisions that turns out to be better than those of a superior player."

Yes, but "luck" in this instance can refer to being lucky that one has a tendency to be optimistic or overconfidence (there's probably a large hertiable aspect to these traits). Being lucky in business can be cast in terms of choosing to focus on a product that consumers turn out to desire--viz., consumer tastes are exogeneous to business skill. Finally, some people have better access to start-up funding than others, for no reason other than luck.

Russell writes:

"Business is like a multi-move game, in which any one decision by an inferior player can turn out right, but it is unlikely that an inferior player will make a sequence of decisions that turns out to be better than those of a superior player."

I believe you've committed the ludic fallacy. One lucky decision in business could have a massive payoff that swamps all the bad decisions. Or vice versa.

Life is really nothing like chess.

Tracy W writes:

Russell - how could one lucky decision in business have a massive payoff that swamps all the bad decisions? My own understand of business is that there's an awful lot of ways to get things wrong, for example, if you stuff up your tax records the tax department can drive you out of business. Or let's say you win a really large account through luck - but you can't deliver on it because your business systems are badly set up. Or you deliver it, but don't manage to extract payment. Or you deliver on it, but a valve fails and oil spills everywhere and you wind up entangled in legal cases for the next 40 years. Or your staff are stealing money from you.

I can believe in one bad or unlucky decision in business swamping all the good decisions, but I have trouble believing in one lucky decision in business swamping all the bad decisions. To succeed in business, I think you need to get a large number of things pretty right.

Tracy W writes:

I'll add that I agree with Arnold's point about the limits of many behavioural economics' experiments. One thing I find interesting is the claim that people aren't motivated by money, which appears to be based on one-off studies where people had already agreed to be in the study. Not paying people for work for longer periods would probably have very different effects on motivation.

Jim Glass writes:
If that one move were the whole game, then the outcome would vary considerably, depending on luck. However, in a sequence of many moves, the better player's chance of coming out ahead gets to be very high.

Business is like a multi-move game, in which any one decision by an inferior player can turn out right, but it is unlikely that an inferior player will make a sequence of decisions that turns out to be better than those of a superior player.

Well, "very high" and "unlikely" can be quantified, and should be for such discussions.

In the model multi-move game, chess, it is quantified by the Elo rating. Signficant random chance always exists even in a "perfect information available" game as this, as shown by the fact that equal-skill players usually beat each other rather than draw (unless they want a draw) and much weaker players have a predictable probability of beating much stronger players over wide strength gaps.

And business is vastly more complicated than that. Not only are there a mulitude of players on each "team" -- working simulataneously both for and against each other -- but nobody has anything like perfect information, and the environment they are competing in is constantly changing in unpredictible ways, which can determine winners and losers all by itself. All this greatly increases the role of chance.

The role of luck in success (and failure) is hugely underestimated in the popular mind and in most after-the-fact analysis and explanation, for a host of reasons.

Take NFL football, a very simple and clearly defined area of competition. More than 40% of game outcomes are random chance (and the percentage is much higher in the playoffs leading to and including the Super Bowl, as teams are more closely matched). But in all the post-game shows and "expert analysis", how often do you hear about all the games determined that week by luck and chance? Never. So what do people believe?

Mike Rulle writes:

The Kahneman posts by Brian and Armold have reminded me why I so much dislike these discussions on "luck" or "randomness", as if these are modern insights which add any value to ordinary common sense. We have free choice, although it is extremely limited. We cannot choose to become giraffes. We cannot choose to be born. We cannot choose our IQ or looks. There are an infinite amount of limitations on our freedom to choose. For all practical purposes, luck is so dominant that it should be viewed as simply a given.

But who does not naturally know this? Ok, many vain people ascribe skill to their luck---after the fact---but what difference does it make? Whatever free choice they had would have been exercised anyway. Determined free decision making and its outcomes may in fact be the residual, but it is ultimately the only thing we can control.

I have trouble understanding this as a meaningful insight.

Claudia Sahm writes:

I agree with your concerns about behavioral economics, which mixes psychology and economics. Someone in grad school explained it to me as: economics is often about average behavior (representative agents) and psychology is often about extreme behavior (gottcha moments, outliers). The mix of two is often a thought provoking mess.

Comments for this entry have been closed
Return to top